Channel Estimation at THz Frequency Using cGAN for Massive MIMO System

In wireless communication, accurate channel estimation is crucial for receiver performance, but predicting the channel in massive MIMO systems is challenging. In this paper, a conditional generative adversarial network (cGAN) is used for accurate channel estimation in a massive MIMO system with one-bit analog-to-digital converters (ADCs). Compared to other deep learning models, cGAN produces high-quality outputs and enables conditional generating by training the generator and discriminator in an adversarial manner to minimize information loss. In this research, we have successfully applied this model to a wide variety of scenarios at 1.4 Terahertz (THz) frequency and achieved satisfactory performance. The simulation results show that in an outdoor 6G massive MIMO system at THz range frequency, an approximate 3 dB gain in estimation error is obtained than in indoor at GHz range frequency.